He will be giving a talk entitled Mandarin wordlikeness megastudies on Monday the 29th at 1:00 PM exactly, in Innis College room 204.
Over the past few years my lab has asked around 200 native speakers of Mandarin to judge the acceptability (“Mandarin-likeness”) of around 10,000 different nonlexical forms, yielding around 700,000 wordlikeness judgments. In megastudies like ours, the large and representative sampling makes it possible to factor out partially confounded lexical variables through multiple regression techniques. Moreover, the corpora of responses generated by megastudies, like corpora of language production, can be freely analyzed for a wide variety of purposes. In this talk I review some of the analyses that we’ve conducted so far on our megastudy corpora. In a “sociolinguistic” analysis, we examined the influence of gender and native language (i.e. Mandarin vs. Taiwan Southern Min). In a “psycholinguistic” analysis, we attempted to tease apart the highly confounded variables of neighborhood density and phonotactic probability by looking at their interactions with reaction time, working memory capacity, and handedness. In two “phonological” analyses, we looked at the controversial syllable position of the Mandarin prevocalic glide and, more ambitiously, at the interaction between lexicality typicality and universal markedness. All of the above analyses used just one of our megastudy corpora, involving monosyllabic test items; analyses of our other megastudies have looked at wordlikeness judgments for disyllabic nonwords and two-character nonwords. We invite the audience to play with our data, freely available on the web, to test their own pet hypotheses, and to join us in the development of our contributor-built cross-linguistic platform for collecting and sharing data from wordlikeness judgment experiments.